No abstract

For diagnosis of coronavirus disease 2019 (COVID-19), a SARS-CoV-2 virus-specific reverse transcriptase polymerase chain reaction (RT-PCR) test is routinely used. However, this test can take up to 2 d to complete, serial testing may be required to rule out the possibility of false negative results and there is currently a shortage of RT-PCR test kits, underscoring the urgent need for alternative methods for rapid and accurate diagnosis of patients with COVID-19. Chest computed tomography (CT) is a valuable component in the evaluation of patients with suspected SARS-CoV-2 infection. Nevertheless, CT alone may have limited negative predictive value for ruling out SARS-CoV-2 infection, as some patients may have normal radiological findings at early stages of the disease. In this study, we used artificial intelligence (AI) algorithms to integrate chest CT findings with clinical symptoms, exposure history and laboratory testing to rapidly diagnose patients who are positive for COVID-19. Among a total of 905 patients tested by real-time RT-PCR assay and next-generation sequencing RT-PCR, 419 (46.3%) tested positive for SARS-CoV-2. In a test set of 279 patients, the AI system achieved an area under the curve of 0.92 and had equal sensitivity as compared to a senior thoracic radiologist. The AI system also improved the detection of patients who were positive for COVID-19 via RT-PCR who presented with normal CT scans, correctly identifying 17 of 25 (68%) patients, whereas radiologists classified all of these patients as COVID-19 negative. When CT scans and associated clinical history are available, the proposed AI system can help to rapidly diagnose COVID-19 patients. The COVID-19 pandemic has rapidly propagated due to widespread person-to-person transmission 1-6. Laboratory confirmation of SARS-CoV-2 is performed with a virus-specific RT-PCR, but the test can take up to 2 d to complete. Chest CT is a valuable component of evaluation and diagnosis in symptomatic patients with

Checkpoint blockade immunotherapies enable the host immune system to recognize and destroy tumor cells1. Their clinical activity has been correlated with activated T-cell recognition of neoantigens, which are tumor-specific, mutated peptides presented on the surface of cancer cells2,3. Here, we present a fitness model for tumors based on immune interactions of neoantigens that predicts response to immunotherapy. Two main factors determine neoantigen fitness: its likelihood of presentation by the major histocompatibility complex (MHC) and its subsequent T-cell recognition. We estimate these two components using a neoantigen’s relative MHC binding affinity and a non-linear dependence on its sequence similarity to known antigens. To describe the evolution of a heterogeneous tumor, we evaluate its fitness as a weighted effect of dominant neoantigens in the tumor’s subclones. Our model predicts survival in anti- CTLA-4 treated melanoma patients4,5 and anti-PD-1 treated lung cancer patients6. Importantly, low-fitness neoantigens identified by our method may be leveraged for developing novel immunotherapies. By using an immune fitness model to study immunotherapy, we reveal broad similarities between the evolution of tumors and rapidly evolving pathogens7–9.

The seasonal human influenza A/H3N2 virus undergoes rapid evolution, which produces significant year-to-year sequence turnover in the population of circulating strains. Adaptive mutations respond to human immune challenge and occur primarily in antigenic epitopes, the antibody-binding domains of the viral surface protein haemagglutinin. Here we develop a fitness model for haemagglutinin that predicts the evolution of the viral population from one year to the next. Two factors are shown to determine the fitness of a strain: adaptive epitope changes and deleterious mutations outside the epitopes. We infer both fitness components for the strains circulating in a given year, using population-genetic data of all previous strains. From fitness and frequency of each strain, we predict the frequency of its descendent strains in the following year. This fitness model maps the adaptive history of influenza A and suggests a principled method for vaccine selection. Our results call for a more comprehensive epidemiology of influenza and other fast-evolving pathogens that integrates antigenic phenotypes with other viral functions coupled by genetic linkage.